binping xiao superconducting surface impedance under radiofrequency field
TRANSCRIPT
Binping XiaoCollider-Accelerator Department, Brookhaven National Lab
Based on the Ph.D dissertation supervised by C. E. Reece and M. J. Kelley
College of William and Mary, Jefferson Lab
How Good Can an SRF Cavity Be?A New First-Principle Calculation of Field-Dependent RF
Surface Impedance of BCS Superconductor
Thin films and new ideas for SRF,
10/7/2014
1
Outline
2
Cavity performance
BCS theory, Mattis-Bardeen (M-B) theory
and extension
Calculation results and explanation
Theory vs. experiments
Summary
J. Bardeen, L. N. Cooper, and J. R. Schrieffer, Physical Review 108,
1175 (1957)
D. C. Mattis and J. Bardeen, Physical Review 111, 412 (1958)
B.P. Xiao, C.E. Reece, M.J. Kelley, Physica C 490 (2013) 26–31
RF surface impedance in cavity
3
A simplified diagram of SRF device.
Introduction to SRF Linac Technology by Rong-li Geng
http://en.wikipedia.org/wiki/Superconducting_Radio_Freque
ncy
current
d
a
a
Zs = Rs + iXs
Cavity Performance
4
H. Pandamsee: “Two Major Open Physics Topic in RF Superconductivity”
1E+09
1E+10
1E+11
0 50 100 150 200
Qu
ali
ty F
acto
r
Peak Magnetic Field [mT]
1.5 GHz 7-cell CEBAF
cavity with 230 μm BCP
230 μm BCP + 34 μm
EP
1.5GHz single cell
CEBAF cavity with 3 h
1400 C baking
All measured at 2.0K
Superheating
Field
• P. Dhakal, G.
Ciovati, G. R.
Myneni, in: IPAC
2012.
• C. E. Reece et al.,
in PAC2005
• C. E. Reece, and
H. Tian, in
LINAC2010
BCS, M-B and extension
5
Each Cooper pair was assumed to have zero net momentum. By minimizing
the free energy one get a single particle distribution function f and Density of
State (DoS), represented by h.
Considering that a single particle scattering from one energy state to
another one, by considering the energy of each state (with one photon added
to either the initial state or the final state.), the probability of initial state and
the possibility of scattering, a scattering form was calculated using f and h.
This scattering form was then applied to the anomalous skin effect to get the
surface impedance.
We assume that all Cooper pairs move with the same velocity Vs, and apply
this assumption to BCS theory to get new f and h.*
These new f and h are applied to get a new scattering form, thus a new
surface impedance.
We average these effects over both RF cycle and depth into the surface to get
the effective surface resistances (Rs) under different fields.**
*J. Bardeen, Reviews of Modern Physics 34 (1962) 667.**I.O. Kulik, V. Palmieri, Particle Accelerators 60 (1998) 257.
Justifications of this extension
6
A mean (phonon) field theory, same as BCS theory, differs from
Eliashberg.
Consider single photon absorption only, same as M-B theory, differs
from de Visser*.
Dealing with moving Cooper pairs, same as Bardeen**.
The way to deal with mean free path, same as M-B theory.
Calculates the “ideal” case, which means the best performance a
cavity can achieve.
A cavity with short mean free path is not necessary to be a “non-
ideal” cavity.
No extra parameters needed except those in M-B theory + residual
How good it can explain experimental results?
*de Visser, P.J., et al., physical review letters, 2014. 112: p.
047004.**Bardeen, J., Reviews of Modern Physics, 1962. 34(4): p.
667.
7
Cooper pair with zero momentumTwo particles can be attractive to each other with
electron-phonon interaction, if they have:
• momentums in opposite direction
• same energy state k (one ↑ the other ↓)
• k in a shell nearby Fermi level kF, with its
thickness determined by Debye energy
Consequence of attraction:
• Bonded particles become Bosons and get
condensed
• Excited particles obey Fermi-Dirac distribution
• A “forbidden zone” appears nearby kF, with its
thickness to be 2𝜟 (energy gap)
• Energy of excited particle (hole below kF and
electron above kF) changes from|εk|=|1/2mvk2-
εF| before condensation to 𝑬𝒌 = 𝜺𝒌𝟐 + 𝜟𝟐
after condensation
A macroscopic quantum effect!
εF=225meVkF
k↑
-k↓ εk=1/2mvk2- εF
2kTD=47.4me
V
(Energies are based on Nb with selected
parameters)
8
Cooper pair with a net momentum
“States with a net current
flow can be obtained by
taking a pairing (k1↑, k2↓)
with k1+k2 =2q, and 2q the
same for all virtual pairs” – quoted from BCS theory
A small net momentum appears
Consequence:
• Energy split appears for ↑ and ↓
• The energy split is angle dependent
• This angle can be any number
𝒗𝑭
𝒗𝒔α
k+q↑
-k+q↓
2q
2εs<0.0048meV
εk+q=1/2m(𝑣𝑘+ 𝑣𝑠)2 - εF= εk + εs + εext
ε-k+q=1/2m(𝑣𝑘- 𝑣𝑠)2 - εF= εk + εs - εext
εext = mvFvs cos α = pFvsx
9
Energy split caused by moving Cooper pairs
BCS theory extension
Before condensation 𝜀𝑘 𝜀−𝒌+𝒒 = 𝜀𝑘 + 𝜀𝑠 − 𝜀𝑒𝑥𝑡 , 𝜀𝒌+𝒒 = 𝜀𝑘 +
𝜀𝑠 + 𝜀𝑒𝑥𝑡
After condensation 𝐸𝑘 = 𝜀𝒌2 + 𝛥2
1
2 𝐸−𝒌+𝒒↓ = 𝐸𝑘 − 𝜀𝑒𝑥𝑡 , 𝐸𝒌+𝒒↑ = 𝐸𝑘 +
𝜀𝑒𝑥𝑡
with 𝐸𝑘 = 𝜀𝒌 + 𝜀𝑠2 + 𝛥2
1
2
0
1
2
3
4
5
-5 0 5
0
1
2
3
4
5
-6 -4 -2 0 2 4 6𝜺𝒌 𝜺𝒌
𝑬𝒌
𝑬−𝒌+𝒒↓
𝑬𝒌+𝒒↑
|𝜺𝒌|
𝒓𝒆𝒇: 𝜺𝑭
Energy needed to break a Cooper pair,
𝜟 for ↑, and the same for ↓ 𝜟 − 𝜺𝒆𝒙𝒕 for ↑, 𝜟 + 𝜺𝒆𝒙𝒕 for ↓
Energy split
f and h modified by moving Cooper pairs
10
Modified density of states and probability of single occupation at T<Tc:
Low field limit density of states
and distribution function
Below EF Above EF Density of states and distribution
function with moving cooper pairs,
angle averaged
-𝟏. 𝟐∆
Plots with 𝑷𝑭𝑽𝒔= 0.4∆and T/Tc=0.97
Density of states and distribution
function with moving cooper
pairs, angle-dependent
-- α = π
-- α = π/2
-- α = 0
-
𝟐∆
1
𝑁(0)
𝑑𝑁(𝐸)
𝑑𝐸=𝑑𝜀
𝑑𝐸
J. P. Turneaure, J. Halbritter, and H. A.
Schwettman, Journal of Superconductivity 4,
341 (1991)
For electron
For hole
For electron
For hole
𝟏. 𝟐∆Also appeared in I.O. Kulik, V.
Palmieri, Particle Accelerators 60
(1998) 257.
Modification of M-B theory by moving Cooper pairs
Scattering happens between
any two k and k’
11
any two k and k’ with q,
𝑬𝒌 + 𝜺𝒆𝒙𝒕 and 𝑬𝒌′ + 𝜺𝒆𝒙𝒕′
BCS/M-B Extension
Net momentum 0 2q
Initial X0
00
(k↑, -k↓)
(k’↑,-k’↓)
(k+q↑, -k+q↓)
(k’+q↑,-k’+q↓)
Final 00
X0
(k↑, -k↓)
(k’↑,-k’↓)
(k+q↑, -k+q↓)
(k’+q↑,-k’+q↓)
Ground(+) or excited(-) +
+
(k↑, -k↓)
(k’↑,-k’↓)
(k+q↑, -k+q↓)
(k’+q↑,-k’+q↓)
Energy difference Wi-Wf Ek↑-Ek’↑ Ek+q↑-Ek’+q↑
Probability of initial state f dependent Modified f dependent
Scattering matrix elements h dependent Modified h dependent
Absorbing/releasing one photon: additional energy difference ±ℏ(ω-is),
s→0
Final expression
• The final expression is a quadruple integration, besides the
integrations in energy and in reciprocal space shown in M-B
theory, the extension has two additional integrations in angles,
related to k and k’.
• The averaging over both RF cycle and depth into the surface
requires two additional integrations.
• A MathematicaTM script was developed to calculate the Rs vs Bpk.
It is slow, but it works.
• No parameter fittings can be done using current script due to the
slow calculation speed.
12
Calculation result and explanation (1)
13
Surface resistance, Rs, (red line) and reactance, Xs, (blue dashed line)
versus Cooper pair velocity and corresponding magnetic field for Nb at 2 K
and 1.5 GHz.
0 50 100 150
584
588
592
596
600
604
608
0
5
10
15
20
25
30
0 200 400 600 800 1000
Magnetic flux density (mT)
Xs
(µΩ
)
Rs
(nΩ
)
vs (m/s)
*M-B
Why decreasing?
Calculation result and explanation (2)
14
E+ħ𝝎E
𝑹 ∝ [𝒇 𝑬 − 𝒇 𝑬 + ℏω ]𝒈𝒅𝑬
The “golden rule” in extreme anomalous limit
and low temperature approximation
Note that 𝑷𝑭𝑽𝒔>>ħ𝝎 could happen, the overlap
between red and purple could be significant.
Net effect: release energy, cause Rs
𝑹 ∝ [𝒇 𝑬𝟏 − 𝒇 𝑬𝟐 ][𝒇 εext +𝒇 −εext ]𝒈(𝒉)𝒅𝑬
In M-B theory, mathematically, the scattering between
any two k and k’ with photon interaction equals to the
scattering between E and E+ħ𝝎.
With moving Cooper pairs, mathematically, the
scattering between any two k and k’ with photon
interaction equals to the scattering between E+εext and
E+ε’ext+ ħ𝝎.
Rs decreasing?• Source: angle between 𝑽𝑭 (any direction) and 𝑽𝒔 cause
energy split with angle dependence.
• Consequence: Energy split and modified single
particle distribution function cause an overall
reduction effect in scattering.
Any E’Any E
Absorb/release a photon
Net effect: release energy, cause Rs
Absorb a photon
𝑷𝑭𝑽𝒔
E+ε’ext+ ħ𝝎E+εext
E+εext E+ε’ext+ ħ𝝎
Absorb
a
photon
Term 1 Term 2 Term 3
0E+00
1E+10
2E+10
3E+10
4E+10
5E+10
6E+10
7E+10
0 20 40 60 80 100 120
Q0
Bpk (mT)
2.0 K, 1.5 GHz - Theory + 1.7 nohm
G1G2 1400C (LG), 1.5 GHz, 2.0 K
2.0 K, 1.3 GHz - Theory + 3.0 nohm
TE1AES005 800C (FG), 1.3 GHz, 2.0 K
TE1AES003 800C (FG), 1.3 GHz, 2.0 K
Calc for: = 32 nm= 40 nm/Tc = 1.85
mfp = 50 nm
Theory vs Experiments
15
Mattis-Bardeen
P. Dhakal, et al., PRST-AB, 2013. 16(4): p.
042001.
A. Grassellino, et al., Supercon. Sci.and Tech.,
2013. 26(10): p. 102001.
No need for extra parameters!
16
More data…Palczewski et al., LINAC2014
About the inconsistency at
the beginning, there are
several possibilities:
1, The choose of
parameters
2, Measurement errors
3, Cavity performance
could be further improved
4, Some facts that are not
considered in this model:
phonon distribution,
multiple photo absorption,
additional non-linear
effects, etc.
“Textbook” values
We actually predicted the
behavior at low temperatures
Eichhorn et al.
17
Exciting? Let’s be honest
0E+00
5E+09
1E+10
2E+10
2E+10
3E+10
3E+10
4E+10
4E+10
5E+10
5E+10
0 5 10 15 20 25 30 35
+5 microns EP
+7 microns EP
+10 microns EP
A. Grassellino et al.
Dhakal et al.
Summary
18
Previous surface impedance calculations are available only for the
low field limit.
A field-dependent derivation of the Mattis-Bardeen theory of SRF
surface impedance has been developed.
The extended range of gradients is treated for the first time.
Without any extra parameters except those from original M-B theory,
field-dependent Rs agreement with experiment with recent heat-
treated/Nb-doping Nb with unusual surface loading is excellent at
different temperatures, with residual resistance to be constant.
The reduction in resistance with increasing field is seen to be an
intrinsic effect.
For type-I, and type-II under Hc1.
What is going to happen between Hc1 and Hc2?
19
Thank you for your attention!
I would like to thank Drs. Ilan Ben-Zvi and
Sergey Belomestnykh for their comments and
suggestions during the preparation of this talk.